Re: Two questions about URIs

> Now, my question is: is there supposed to be any particular
> relationship between the referent of such a URI, considered
> as a name, and the file that it locates in the global file network?

This is under heavy debate. It is often regarded that the utopian vision of
URIs is that they are simply identifiers for resources, and that the
resolution characteristics of the URI are something that the user decides.
TimBL has stated that an HTTP URI is not some magic instruction to open up
a TCP connection on port 80 to some server and use the HTTP protocol to
retrieve a file, but is instead a generic identifier of which any
resolution method can be applied [in the XML 2000 talk]. Contrariwise, he
states that HTTP URIs necessarily identify generic documents [1]. Seems
like a contradiction, but really IMO it's just two sides of the same coin,
as I shall explain in a minute. Roy Fielding has often argued that HTTP
URIs identify a range of resources much wider than generic document and
data types, and indeed HTTP request codes and work conducted on
HTTP-similar protocols such as HTTPS suggest that HTTP URIs do not
necessarily identify documents or files.

The whole field in this regard is subjective. It is clear that practically,
people expect certain URIs to have a particular context-of-use, that is, in
Webpages, as something that can be simply entered into a browser. The
ambiguity over what a URI identifies comes from the fact that the mapping
from the identifier to the resource is decided only by the context of its
use.

Some examples, moving from large scale to small scale [non exhaustive]:-

http://example.org/
* References the HTTP server of the example.org domain
* References the concept that is denoted by the network entity (go to
example.org for some great widgets!)
* References the network entity (as Fred wrote on example.org, "widgets are
cool")

The fact of the matter is that it doesn't matter too much what context an
identifier is used for, so long as people understand that identifiers used
in different contexts have different levels of persistence (on the above,
you'd expect it to always reference the example.org HTTP server, nearly
always reference the concept that example.org, and perhaps change often
enough that citing material from it is a temporary thing). When the context
of use of an identifier is clear, then they can often be used more
effectively and persistently, and that's the thinking behind URNs.

Cheers,

[1] http://www.w3.org/DesignIssues/Fragment

--
Kindest Regards,
Sean B. Palmer
@prefix : <http://webns.net/roughterms/> .
:Sean :hasHomepage <http://purl.org/net/sbp/> .

Received on Tuesday, 16 October 2001 16:10:23 UTC